Assessment of aggregated indicators of sustainability using PCA: the case of apple trade in Spain

Assessment of aggregated indicators of sustainability using PCA:

the case of apple trade in Spain
J. Soler-Rovira1 and P. Soler-Rovira2
1
Departamento de Produccion Vegetal: Fitotecnia, Escuela Universitaria de Ingenieria Tecnica
Agricola, Universidad Politecnica de Madrid, Ciudad Universitaria s/n, 28040-Madrid, Spain,
jose.soler@upm.es
2
Instituto de Ciencias Agrarias, Centro de Ciencias Medioambientales, C.S.I.C., Serrano 115 dpdo.,
28006-Madrid, Spain.

Keywords:, ecological footprint, economic impact, environmental indicators, life cycle assessment,
multivariate analysis, social impact, sustainability.

Abstract
Environmental, economic and social impacts of intensive agricultural production, but also those
regarding to international trade of fresh apples in Spain, were studied by the multivariate statistical
method of principal components analysis (PCA). Environmental indicators were developed for 36
countries using life cycle analysis of apple cultivation and transport, weighting the results on a global
or local scale. Economic and social indicators were also calculated considering macro and
microeconomic aspects and also farm or society characteristics. PCA was applied to each set of
indicators and aggregated indices were computed for each dimension of sustainability with the results
of the analysis. The selected indicators explained with good agreement the differences in sustainability
between countries and the synthetic indices ranked them all. Some of them showed a high relative
sustainability, while other presented low values, due to low environmental, economic or social
sustainability values of the aggregated indices.

Introduction
In recent years, international trade of fresh fruits and vegetables has also increased due to market
liberalisation and technical development of agricultural practices, conservation processes and transport
facilities. There are a variety of complex environmental, economic and social impacts regarding
international trade. Those impacts arise at either global or local scales, pertaining to issues as energy
consumption, emission of pollutants, degradation of natural resources, land-use changes, etc. By the
other hand, economic growth takes place in exporting countries, but wealth often shows and unequal
distribution between the populations (Wurtenberger et al., 2006). Consequently, there is a need to
assess the environmental, economic and social impacts of intensive agricultural production but also
those regarding to international trade.
Spain is a big fruit producer with approximately 13 million tonnes in 1,300,000 ha of cultivated land,
which represents almost 30% of the harvested area of fruits in the EU (FAO, 2008). The trade flows of
agricultural products with other countries show a balanced state since the imports equals the exports.
The main exported agricultural product is the fresh fruit (4,300 million euro in 2005), but Spain also
imports fruits for 1,300 million euro (MAPA, 2006). Apple production and trade is a good case study
because it accounts for 14% of total fruit consumption in Spain, and 29% of this quantity is imported
from other countries (MAPA, 2006). Spanish imports of apple have been increased in the last years
and the main origins are France, Italy, Germany and Portugal in the EU, and Chile and Argentina in
South America (FAO, 2008). Imports from France have decreased in the period 2000-05, but the other
origins show an increasing trend, except Portugal that shows a flat evolution.
The sustainability of this apple trade may be assessed by a number of indicators that reveal the impacts
of their cultivation and transport from producer countries to the Spanish market. These indicators
should be useful for policy-makers at the roundtables where trends are monitored and sustainable trade
policies are introduced and evaluated. Even more, they can be helpful product information for
consumers and their associations towards a consumption trend that accomplish a set of sustainability-

Proc. of the 6th Int. Conf. on LCA in the Agri-Food Sector, Zurich, November 12–14, 2008 page 133 of 414
Assessment of aggregated indicators of sustainability using PCA: the case of apple trade in Spain

sensitive criteria (Levitan, 2000). Indicators should be synthesized into an appropriate indicator that
contains a lot of information but, at the same time, it is easy to understand by the end-users (policy-
makers, consumers, etc.). Aggregated indicators help to communicate the information succinctly and
make easier to distinguish patterns in the data by formalizing the aggregation process that is often
done implicitly, subjectively and intuitively (Jollands et al., 2004). Nevertheless, it is necessary to
consider the potential limitations of the synthetic indicators since they may mask and simplify the
complexity of environmental, social and economic systems.
In the development of aggregated indicators problems arise when the indices that build up the
indicator and the weights of each index have to be selected. Therefore, principal components analysis
(PCA) can be used as an objective approach to choose the indices that show higher variability within
the studied observations and to set the weights as a function of the explained variance (Jollands et al.,
2004). However, PCA is limited to ex-post analysis and it is not an appropriate tool for prospective
investigations. Additionally, this analysis allows making an internal sustainability evaluation between
countries, giving a relative value of sustainability.
PCA have been used in several studies that include large sets of data, i.e. ecology and water quality,
landscape characterization, pesticide screening or food quality. PCA has been applied to select proper
and representative variables that could explain the variability included in the original data. The
usefulness of PCA have been demonstrated to select environmental (Yu et al., 1998), energy intensity
(Bernard and Cote, 2005), eco-efficiency (Jollands et al., 2004) and agri-environmental indicators
(Soler-Rovira and Arroyo-Sanz, 2003; Soler-Rovira and Arroyo-Sanz, 2004). For example, the latter
authors selected nutrient management indicators and classified the Spanish provinces and districts
applying cluster analysis to the results from the PCA. Finally, synthetic indicators have been
aggregated by PCA for data obtained in studies on sustainable agricultural systems (Sands and
Podmore, 2000), irrigation schemes (Rodriguez-Diaz et al., 2008), poverty and human development
(Antony and Visweswara Rao, 2007) and sustainable development and environmental quality (Castro,
2002; Jha and Murthy, 2003; Escobar, 2006). However, the methodology used to build up the
aggregated indicator differs between authors and none of them have applied the aggregation with PCA
combined with life cycle analysis on sustainability of production and trade.
The aim of this work is to assess the sustainability of apple production and trade flows in Spain by the
development of aggregated indexes obtained by multivariate analysis (PCA) of individual indicators
(economic, social and environmental).

Method
The methodology used is PCA, so first of all this multivariate statistical tool is briefly described in this
section. Secondly, the characteristics of apple trade in Spain are evaluated and the main apple
producing countries are selected. The next steps are to choose a set of indicators that can be used to
characterize the sustainability of the environmental, economic and social dimensions of apple trade.
Lastly, all the selected indicators are synthesized in a aggregated index of sustainability. All these
steps are extensively described below.

A brief description of principal components analysis

PCA is a statistical multivariate methodology used to study large sets of data. This method reproduces
a great proportion of variance among a big number of variables by using a small number of new
variables called principal components (PCs). The PCs are linear combinations of the original
variables, and the analysis of multidimensional data is simplified when these are correlated (Judez,
1989). The first PC explains maximum variance between data, while the second component is a new
combination of the original variables being orthogonal to the first component and explaining the
second largest value of variation among observations, and so forth. The absorption of variance in each
component is computed with the so-called eigenvalues. One property of the PCs is that they are
uncorrelated between them, and then each component is measuring a different dimension in the data.

Proc. of the 6th Int. Conf. on LCA in the Agri-Food Sector, Zurich, November 12–14, 2008 page 134 of 414
Assessment of aggregated indicators of sustainability using PCA: the case of apple trade in Spain

High absolute values of loadings of the variables (i.e. indicators) on the PCs imply that the indicator
has a large bearing on the creation of that component. Thus, the most important indicators in each
component, that best explain variance; will also be more useful in explaining variability between
observations (i.e. countries). Each component will be a linear combination of indicators (variables)
multiplied by their loadings on that component. Observations (countries) will have coordinates in each
axis or component, computed with the standardized value of each variable (zero mean and unit
variance) for that observation using the linear combination of variables with PCs obtained in the
analysis.

Apple trade in Spain

Twenty most important apple exporters of the world and other 16 countries that have exported apples
to Spain in the last 10 years have been selected as the observations set. Data of apple exports and
imports are from FAO (2008). The 36 selected countries are shown in Tab. 3.

Environmental indicators
The environmental dimension was analysed considering the crop production and the transport of the
apples. Agricultural practices were assessed searching information about fertilization, irrigation and
yield of apple orchards, i.e. FAOSTAT (FAO, 2008), International Fertilizer Industry Association
(IFA, 2008) and Water Footprint of Nations (Waterfootprint, 2008). The agricultural impact was
calculated for 1 hectare (ha) of orchard and for 1 kilogram (kg) of fresh apples (just dividing by yield).
Irrigation impact was considered as the water requirements of the crop for one year (Waterfootprint,
2008) in m3 per ha or per kg of apple. Fertilization impact was assessed by computing emissions and
inputs during manufacturing of fertilizers, so data per kg N, kg P2O5 and kg K2O manufactured were
used. A nutrient balance was carried out in apple orchards, considering atmospheric N emissions from
fertilizers (NH3, N2O and NO) with EMEP methodology (EEA, 2004) and nitrate leaching as the mean
of a constant value of 16% of N inputs in fertilizers (Neilsen and Neilsen, 2002) and the result of the
balance Nleaching= Nfertilizers-Ncrop uptake - Ngaseous emissions, when it was positive. The inventory of transport
was done with the distance from production zones to Madrid (Spain) and using lorry and ship
emissions per t km transported. Road transport by lorry was considered from countries in continental
Europe and distance was computed by data from ViaMichelin (2008). Sea transport by ship was
considered from the other countries, computing distance from the main port of the country to
Algeciras or Valencia in Spain (the two main ports for fruit trade), via Panama, Suez or Gibraltar (Sea
Distances, 2008). Lorry transport from the ports to Madrid was also taken into account. Life cycle
analysis of transport data was done for 1 kg of fresh apples from each country.
Ten impact categories were considered: global warming, acidification, eutrophication, human toxicity,
ecotoxicity in fresh water, photochemical oxidants formation, energy use, water resources use, abiotic
resources depletion and land use. Characterization factors for each category were use from CML-IA
(2004). World in 1995 normalization factors were used (Van den Berg et al., 1995; Huijbregts et al.,
2003; CML-IA, 2004).
Normalized values of the LC analysis of each impact category were added up for crop production
(LCAcrop indicator) and for transport (LCAtransport indicator), and the sum of those two was an
overall potential environmental impact indicator (LCAtotal). Other two indicators were calculated
considering a local and a global geographical scale. The impact of apple production over local
population and ecosystems was calculated per ha of cultivated land, considering that the main impact
categories were toxicity for human population, depletion and pollution of water resources and land use
and occupation for agriculture. A multi-criteria analysis was carried out using analytical hierarchy
process (AHP) (Saaty, 1990). The ten studied impact categories were ranked in a sequence from more
to less relative importance at local scale: Human toxicity = water resources use = eutrophication =
ecotoxicity fresh water = land use > acidification = photochemical oxidants formation > energy use =
abiotic resources depletion > global warming. Based on these assumptions, the respective weights for
each of the ten impact categories were calculated according to the AHP procedure. These weights
were applied to the crop LC analysis and a local impact indicator was considered (LCAlocal).

Proc. of the 6th Int. Conf. on LCA in the Agri-Food Sector, Zurich, November 12–14, 2008 page 135 of 414
Assessment of aggregated indicators of sustainability using PCA: the case of apple trade in Spain

The global impact of apple production and trade was calculated per kg of apple, considering that the
main impact categories were climate change, energy use and depletion of natural resources. An AHP
analysis was carried out sorting the ten impact categories in a sequence from more to less relative
importance at global scale: Global warming > energy use = abiotic resources depletion = land use >
water resources use = eutrophication = ecotoxicity fresh water > acidification > human toxicity >
photochemical oxidants formation. The weights were calculated as for the local scale and they were
applied to the crop and transport LC analysis and a global impact indicator was considered
(LCAglobal).
Other environmental indicators were also calculated pertaining to particular aspects of environmental
impacts, as productive land requirements, use of resources or emissions during the apple life-cycle.
Ecological footprint was determined considering the yield of the orchards and the CO2 emitted during
fertilizers manufacturing and apple transport. Arable land and sink forest land for CO2 were calculated
and equivalence factors (Wackernagel et al., 1999) were applied to determine the ecological footprint,
i.e. m2 of land required per kg of apples. Carbon footprint was computed as the kg of CO2 equivalent
per kg of apples emitted during cultivation and transport. Water footprint was calculated considering
the yield and the water requirements in each country (m3 ha-1) in L of water per kg of apples. Energy
footprint was determined as the energy used in fertilizer manufacturing and apple transport (MJ kg-1).
Reactive nitrogen released to the biosphere during fertilization and NOx emitted in fertilizer industries
and apple transport were also calculated (g N kg-1).
A synthetic environmental indicator was calculated by PCA using a matrix of 18 variables x 36
countries. The initial set of environmental indicators included a large set of variables in order to firstly
investigate which of them showed higher variability within the studied observations and correlation
within them, that is strength of PCA, although some of them should explain redundant information.
Thus, fertilization rates (kgN/ha, kgP/ha and kgK/ha), fertilizers per unit of apple produced (kgN/kg,
kgP/kg and kgK/kg), water requirements (m3/ha), water footprint (L/kg), transport distance (km),
ecological footprint, energy footprint, carbon footprint, reactive nitrogen, and LCA values (LCAcrop,
LCAtransport, LCAtotal, LCAglobal and LCAlocal) were included. Before developing the PCA, all
the variables were signed as positive or negative in order to make them unidirectional (Jha and
Murthy, 2003). PCA was performed with STATGRAPHICS software, standardizing data to zero mean
and unit variance. Eigenvalues and the amount of variance explained by each principal component
(PC) were calculated. The number of components retained in the analysis was assessed by Cattel’s
scree plot, which indicates that we should retain i components because, after the i+1 component, the
plot becomes flat, corresponding to eigenvalues lower than one. The value of the eigenvectors and
loadings of variables with PCs were computed. Coordinates of each country with each axis were
determined. The aggregation of data into a single environmental sustainability index was calculated as:
j

∑ Fki λk
PCAenvironmental (i ) = k =1
j
i = 1,...,36(countries) [1]
∑k =1
λk

Where, Fki is the coordinate of the country i in the component k (and j components are retained) and λk
is the eigenvalue of the component k. This index should give information about the relative value of
environmental sustainability between the studied countries, taking into account that LCA gives an
estimation of potential impacts.

Economic indicators
Economic dimension was analysed considering micro and macro economic aspects of apple trade.
Micro economic level was studied at farm scale, so productivity, yield stability and yield sustainability
were calculated as fundamental properties of farming systems (Marten, 1988). These indicators were
computed as Tab. 1 shows. Macro economic level was assessed by nine indicators. One studied aspect
was the positive effects of exportation of agricultural products as the returns obtained by apple
exports, with a high market share and competitiveness. On the other hand, some negative aspects

Proc. of the 6th Int. Conf. on LCA in the Agri-Food Sector, Zurich, November 12–14, 2008 page 136 of 414
Assessment of aggregated indicators of sustainability using PCA: the case of apple trade in Spain

would arise as market oligopoly with an export-oriented farming with apple cultivation as a
monoculture. Other negative factors could be the decay of exportation prices (Barriga, 2003) or the
decreasing and abandonment of apple cultivated area. Other interesting indicator is called
globalization in the sense that in some countries a great volume of imports and, at the same time,
exports of apples exists, so the national market is decidedly open to the global market. All the
statistical data were obtained from FAO (2008).
A synthetic economic index was calculated by PCA using a matrix of 12 variables x 36 countries (Tab.
1). All the variables were signed as positive or negative in order to make them unidirectional (Jha and
Murthy, 2003). PCA was performed as described above and the aggregation of data into a single
economic sustainability index (PCAeconomic) was computed as in equation [1].

Tab. 1: Initial set of economic and social indicators used in the analysis.
Economic dimension
Indicator Calculation
Productivity Average yield of apple in the period 1996-2003.
Yield stability Coefficient of variation of apple yield during 1996-2003.
Yield sustainability si/Yi; where: si: slope of yield over the period 1996-2003 in country i.
Yi: average yield in country i during that period.
Exports value (AEVi/AgEVi)*100; where: AEVi: average apple exports value in country i in 2000-05
period. AgEVi: average agricultural exports value in country i in 2000-05 period.
Market share (AECi/AEW)*100; where: AECi: apple exports in country i over the period 1996-2004.
AEW: total apple exports during that period in the world.
Competitiveness Esi/Ei; where: Esi: slope of apple exports in country i over the period 1996-2004.
Ei: average exports in country i during that period.
Oligopoly MSi-AMS; where: MSi: market share of country i.
AMS: average market share of each country if all world exports where fairly distributed.
Export-oriented (AEAi/TAAi)*100; where: AEAi: area of exported apples in country i in the period 1996-
farming 2004. TAAi: total area of apple in country i in the period 1996-2004.
Monoculture (AAi/TFAi)*100; where: AAi: area of cultivated apples in country i in the period 1996-
2004. TFAi: total fruit cultivated area in country i in the period 1996-2004.
Exports price Slope of apple exports prices over the period 1996-2004 in each country.
Abandonment Slope of apple cultivated area over the period 1986-2006 in each country.
apple area
Globalization [(AIi+AEi)/APi)]*100; where: AIi: apple imports in country i in the period 1996-2004.
AEi: apple exports in country i in the period 1996-2004.
APi: apple production in country i in the period 1996-2004.
Social dimension
Indicator Calculation
Income stability Coefficient of variation of orchard income (yield x price) during 1996-2003.
Income trend Slope of income over the period 1996-2003.
International justice APPi-APPav; where: APPi : apple producer prices in country i during 1996-2003.
APPav: apple producer prices of the major 20 exporting countries during that period.
Market-farmer (APPi/ACPSp)*100; where: APPi: apple producer price in country i during 1996-2003.
equity ACPSp: apple consumer price in Spain in 2007/08.
Fruit deficit FVCi-FVCRWHO; where: FVCi: fruits and vegetables consumption per capita in country i
in the year 2003. FVCWHO: fruits and vegetables consumption recommended by WHO
(400 g day-1).
Fruit diversity Shannon index of fruit consumption per capita in year 2003.
Food waste AWi/ASi; where: AWi: apple waste in country i in year 2003.
ASi: apple supply in country i in year 2003.
Own supply ASi/FVCi; where: ASi: apple supply per capita in country i in year 2003.
FVCi: fruits and vegetables consumption per capita in country i in year 2003.

Proc. of the 6th Int. Conf. on LCA in the Agri-Food Sector, Zurich, November 12–14, 2008 page 137 of 414
Assessment of aggregated indicators of sustainability using PCA: the case of apple trade in Spain

Social indicators
Social dimension was analysed considering farm level and society level. Farm level was assessed with
four indicators and, first of all, farm income stability and income trend were calculated (Tab. 1).
Market-farmer equity calculates the percentage of the final price (paid by consumer for 1 kg of apples)
that farmers receive. International justice was also computed as the difference in revenues between
apple in one country and average revenues from apples of the 20 major apple exporting countries
(Wurtenberger et al., 2006). Social aspects are related to fruits and vegetables deficit in the diet from
the minimum recommended by World Health Organization (i.e. 400 g capita-1 day-1) and diversity of
types of consumed fruits. Other social aspects are food waste and own supply of apples. All the
statistical data were obtained from FAO (2008), except apple price paid by consumers in Spain
(MAPA, 2008).
A synthetic social index was calculated by PCA using a matrix of 8 variables x 36 countries (Tab. 1).
All the variables were signed as positive or negative in order to make them unidirectional (Jha and
Murthy, 2003). PCA was performed as described above and the aggregation of data into a single social
sustainability index (PCAsocial) was computed as in equation [1].

Sustainability index
A synthetic sustainability index was built up considering environmental, economic and social
indicators studied in the previous sections. Eleven environmental indicators were selected from the
initial set of 18, considering those that showed a high correlation coefficient with the PCAenvironmental
aggregated index, discarding also those that gave redundant information (e.g. fertilization rates or
LCAtotal indicator). The same procedure was carried out to select 9 economic and 8 social indicators.
PCA was performed using a matrix of 28 variables x 36 countries. Aggregation was done as showed in
equation [1] and a PCAsustainability index was computed.

Results and discussion

Principal components analysis for the environmental indicators is shown in Tab. 2. Five principal
components were retained and they explained 93.5% of the total variance of the data. The first
component (PC1) is highly correlated with five indicators that describe ecological footprint (EF),
water footprint and LCA results for crop, total system and global scale. These indicators are related to
yield and water consumption of apple orchards. The second PC is correlated with LCA in transport
and related indicators as carbon and energy footprints and distance covered. PC3 is correlated with
nitrogen and potassium fertilization, PC4 with impacts at a local level (per ha) and PC5 with
phosphorus fertilization.
These indicators will explain with good agreement the differences in environmental sustainability
between countries, and the coordinates of each country with each component will built up the
synthetic environmental index, weighted with the eigenvalues of each component. The resulting index
for each country and the corresponding ranking between all, are shown in Tab. 3. Twenty countries
show positive values, thus higher than the mean (that is zero). The other 16 countries show negative
values. France, Netherlands, Belgium and Switzerland are in the best positions in the relative
hierarchy of environmental sustainability, while China, Cyprus, Iran and Korea are in the lower part of
the ranking.
Another five principal components were retained in the analysis of the economic dimension (Tab. 2).
The first PC would be defined as international trade, as it is positively correlated with market share
and the subsequent revenues from exports, and negatively with market oligopoly. The economic
sustainability index is positive in 17 countries and the other 19 show values under the mean. The first
positions in the relative ranking are held by Chile, USA, Italy and France, while Finland, Latvia,
Cyprus and Morocco are located in the last positions.
Social indicators were explained by four PCs. The first PC shows high loadings with market-farmer
equity and international justice related to apple prices, and with food waste. The second PC is highly
correlated with farm income and fruit deficit in population’s diet. Regarding the aggregated social

Proc. of the 6th Int. Conf. on LCA in the Agri-Food Sector, Zurich, November 12–14, 2008 page 138 of 414
Assessment of aggregated indicators of sustainability using PCA: the case of apple trade in Spain

index ranks Cyprus, UK and Switzerland in the first positions of the 25 countries with positive value,
while 11 show negative values as Moldova, Chile and Latvia that show the lowest values.
Tab. 2: Principal component analysis for environmental, economic, social and sustainability indices.
Environmental indicators
PCs retained PC1 PC2 PC3 PC4 PC5
Eigenvalues 7.34 4.41 3.00 1.84 1.18
Variance absorption 38.6 61.8 77.6 87.3 93.5
(%)
EF (0.95) LCAtransport kgK/ha (0.92) LCAlocal kgP/ha (0.88)
Correlated indicators LCAcrop (0.94) (0.99) kgK/kg (0.91) (0.96) kgP/kg (0.85)
(loadings) WaterF (0.94) EnergyF (0.96) kgN/kg (0.73) m3/ha
LCAtotal (0.93) CarbonF (0.95) kgN/ha (0.67) (0.96)
LCAglobal (0.91) Distance (0.87) Nreactive (0.63)
Economic indicators
PCs retained PC1 PC2 PC3 PC4 PC5
Eigenvalues 3.44 1.90 1.60 1.30 1.10
Var. absorption (%) 28.6 44.5 57.8 68.6 77.4
Oligopoly Yield sustainability Exports price Ex. Monoculture
Correlated indicators (-0.95) (0.82) (0.80) oriented (0.91)
(loadings) Market share Globalization (0.82) Abandonment (-0.83)
(0.95) Yield stability (-0.69) Productivity
Exports value (0.60) Competitiveness (0.82)
(0.60) (0.66)
Social indicators
PCs retained PC1 PC2 PC3 PC4
Eigenvalues 2.18 1.70 1.18 1.00
Var. absorption (%) 27.3 48.5 63.2 75.8
Correlated indicators Equity (-0.95) Income trend (0.82) Own supply (0.92) Fruit diversity
(loadings) Int. justice (0.72) Fruit deficit (0.60) (0.88)
Food waste (0.50) Income stability (0.56)
Sustainability index
PCs retained PC1 PC2 PC3 PC4 PC5 PC6 PC7
Eigenvalues 6.63 5.45 3.60 2.90 1.80 1.48 1.16
Var. ab. (%) 23.7 43.1 55.9 66.3 72.8 78.0 82.2
LCAcrop (0.94) LCAtransport LCAlocal Income trend Market Exports Fruit
Correlated EF (0.93) (0.96) (0.92) (0.85) share value deficit
indicators WaterF (0.93) CarbonF m3/ha Yield sust. (0.91) (-0.83) (0.74)
(loadings) LCAglobal (0.89) (0.96) (0.92) (0.84) Oligopoly Food
Productivity (0.80) EnergyF Fruit Globalization (-0.91) waste
Int. justice (0.64) (0.94) diversity (0.77) (0.81)
Income st. (0.62) Distance (-0.46) Equity Export
Yield st. (0.58) (0.85) (-0.57) oriented
Exp. price (0.58) Nreactive farming
(0.62) (0.50)

The results of the PCA for the 28 selected indicators (11+9+8) in order to develop a sustainability
index are shown in Tab. 2. Seven principal components were retained and they explained 82.2% of the
total variance of the original data. The first component (PC1) shows high loadings with a combination
of environmental, economic and social indicators. The environmental ones were ecological and water
footprints and LCA results for crop and global scale, so the main environmental issues were related to
water and land use per kg of apples produced. The economic dimension was explained by two
microeconomic indicators (yield productivity and stability) and one macroeconomic (trend of export
prices).. Social sustainability was related to apple price indicators (international justice and farm
income). The second PC is only correlated with environmental indicators related to apple transport, as
LCA added values of transport, the distance covered, and the related energy consumed and equivalent
carbon emitted. Moreover, reactive nitrogen emissions are captured by this second PC, although they

Proc. of the 6th Int. Conf. on LCA in the Agri-Food Sector, Zurich, November 12–14, 2008 page 139 of 414
Assessment of aggregated indicators of sustainability using PCA: the case of apple trade in Spain

consider the agricultural and transport phases. The third component is positively correlated with
impact al local scale, mainly produced by water use per ha, and negatively with diversity in fruit diet.
The last four PCs capture economic and social issues. These indicators will explain with good
agreement the differences in overall sustainability between countries. The coordinates of each country
with each component will built up the synthetic sustainability index for that country, weighted with the
eigenvalues of each component (Tab. 3). Twenty countries show positive values of the synthetic
index, and the other 16 are below the mean and show negative values in the relative ranking of
sustainability performed by the analysis.
Tab. 3: Values of the aggregated indices for sustainability and environmental, economic and social
dimensions of the 36 countries studied. Relative ranking of each country is also shown.
Country PCAenviron Ranking PCAecon Ranking PCAsocial Ranking PCAsustainability Ranking
Argentina -1.494 27 0.427 14 -0.980 31 -0.671 22
Austria 3.007 6 0.757 11 0.651 11 1.976 7
Belgium 3.523 3 1.237 9 0.400 15 2.416 4
Brazil -0.008 21 0.367 15 -0.914 30 -0.846 24
Canada 1.819 13 -0.138 19 0.294 18 1.738 10
Chile -0.620 25 2.789 1 -2.124 35 -1.151 26
China -10.708 36 1.512 5 -0.733 29 -3.599 35
Cyprus -9.614 35 -1.839 34 1.240 1 -3.476 33
Czech R. 2.875 7 -0.366 22 0.112 22 0.849 17
Denmark 1.682 15 -1.060 31 0.773 7 1.147 12
Finland -1.799 28 -4.149 36 0.285 19 -2.314 32
France 3.875 1 1.987 4 0.147 21 2.769 1
Germany 2.061 12 0.708 12 0.373 16 0.957 16
Greece 1.428 18 -0.722 27 0.853 6 0.754 18
Hungary -0.451 23 -0.403 23 -1.088 32 -0.684 23
Iran -4.587 34 0.315 16 0.544 14 -2.147 30
Ireland 3.131 5 -0.524 24 0.162 20 2.183 5
Italy 2.070 11 2.284 3 0.547 13 1.934 8
R. Korea -3.798 33 -0.189 20 0.667 10 -2.187 31
Latvia -1.170 26 -3.804 35 -1.873 34 -3.510 34
Moldova -2.937 32 -1.425 32 -3.063 36 -4.252 36
Morocco -2.906 31 -1.604 33 0.065 23 -1.751 27
Netherlands 3.619 2 0.838 10 0.696 9 2.513 3
N.Zealand -0.546 24 1.288 7 0.013 25 -0.924 25
Poland -0.101 22 1.389 6 -1.744 33 0.221 20
Portugal 0.074 20 -0.548 25 0.623 12 0.646 19
Slovakia 2.138 10 -0.826 30 -0.207 27 -0.054 21
Slovenia 2.861 8 0.147 17 0.029 24 1.423 11
S. Africa -2.569 30 1.240 8 -0.542 28 -1.985 29
Spain 1.690 14 -0.131 18 0.312 17 1.075 14
Sweden 0.112 19 -0.822 29 0.862 5 1.064 15
Switzerland 3.508 4 0.628 13 0.966 3 2.695 2
Turkey 1.596 16 -0.286 21 0.744 8 1.093 13
UK 2.671 9 -0.817 28 0.970 2 1.810 9
USA 1.437 17 2.357 2 0.950 4 2.096 6
Uruguay -1.869 29 -0.614 26 -0.008 26 -1.808 28

Within the first 19 countries we can slightly separate two different groups. The first group include nine
countries that show a positive value of the sustainability index and also show this positive value in the
three previous computed indices pertaining environmental, economic and social sustainability. These
countries show, in general, lower environmental impact in a global scale, a high productivity and a
good justice for apple prices within the global market, although they have a tendency to monopolize it.
This group includes USA and eight European countries: France, Switzerland, The Netherlands,
Belgium, Austria, Italy, Slovenia and Germany. The second group is characterized by positive values
of the aggregated sustainability index; they have a socio-environmental sustainability in the relative
hierarchy ranked in the analysis, because only the environmental and social indices are positive and

Proc. of the 6th Int. Conf. on LCA in the Agri-Food Sector, Zurich, November 12–14, 2008 page 140 of 414
Assessment of aggregated indicators of sustainability using PCA: the case of apple trade in Spain

the economic index is negative. The environmental dimension is characterized by relative low
environmental impact, particularly energy and carbon footprints, and high social sustainability,
especially with regard to prices, income and waste indicators. The worse side is the low economic
sustainability, represented by productivity and market share indicators. This group includes Canada,
Turkey and eight European countries (Ireland, UK, Denmark, Spain, Sweden, Czech Republic, Greece
and Portugal).
Another seventeen countries are in the low zones of the relative sustainability ranking established by
the PC analysis. New Zealand and Iran show a positive socioeconomic sustainability index, with an
important weight of apples in agricultural exports and relative good values for income and fruit
consumption indicators. However, the environmental sustainability index is negative due to impacts at
global scale and during the transport stage.
Argentina, Brazil, Chile, China and South Africa show a negative value of the sustainability index and
reach only a positive value of the economic index. Environmental impact is relatively high for
agricultural production and, above all, the transport phase due to the large distances to Spain. This
impact is important at both global and local scale. Social aspects are eroded by low producer prices
and deficit of fruit consumption. The economic advantages that show these countries are the market
share for apples, but they tend to control it. Poland would be included with these economic sustainable
countries, although it shows a positive sustainability index.
Slovakia is characterized by a positive environmental sustainability index due to global and local
relative low impacts. However, economic aspects are shorted by low yield sustainability; and social
dimension is deficient with regard to income trend of farmers and fruit consumption of the population.
Positive social sustainability index is achieved by Cyprus, Finland, Republic of Korea and Morocco
due to a relative high equity and justice in apple prices. Economic sustainability indicators as
productivity and apple monoculture should be improved. The environmental impact at a global scale is
another bad indicator.
The less sustainable indices are shown by Hungary, Latvia, Moldova and Uruguay. The three
dimensions of sustainability show negative values and they are in the lower part of the sustainability
index ranking. Productivity and monoculture should be improved in the economic dimension. Equity
and justice of producer prices should be enhanced and deficit of fruits and vegetables consumption
should be reduced. Environmental impact at a global scale should decrease, and ecological and water
footprints should be improved.

Conclusion
Principal components analysis is a good statistical tool to develop aggregated indicators in order to
asses the sustainability of apple production and trade flows in Spain. This multivariate analysis can be
used as an objective approach to select the most important indicators regarding economic, social and
environmental aspects of apple production and trade. The aggregation of data yields a single index
easy to understand and that contains a lot of information, and allows to make a ranking between
studied countries. Then, the sustainability of apple trade may be assessed by synthetic indices and
strengths and weaknesses of each country may be discerned, and improvements may be suggested by
studying individual indicators. The results for the main producing countries of apples imported in
Spain show that France and Italy have a high sustainability index, Spain and Portugal just have
positive values for the social and environmental aspects, while Argentina and Chile showed only
positive values for economic sustainability.

References
Antony, G.M., Visweswara Rao, J.2007. A composite index to explain variations in poverty, health, nutritional
status and standard of living: Use of multivariate statistical methods. Public Health, 121: 578-587.
Barriga Montes, I.J. 2003. Tendencia en el mercado de la manzana en los últimos 20 años. Universidad de Talca,
Facultad de Ciencias Agrarias, Escuela de Agronomia, Talca, Chile, 71 pp.

Proc. of the 6th Int. Conf. on LCA in the Agri-Food Sector, Zurich, November 12–14, 2008 page 141 of 414
Assessment of aggregated indicators of sustainability using PCA: the case of apple trade in Spain

Bernard, J.T., Cote, B. 2005. The measurement of the energy intensity of manufacturing industries: a principal
components analysis. Energy Policy, 33: 221-233.
Castro, J.M. 2002. Indicadores de desarrollo sostenible urbano. Una aplicación para Andalucia. Ph.D. Thesis,
University of Malaga, Spain, 540 pp.
CML-IA. 2004. Institute of Environmental Science, Leiden University. In:
http://www.leidenuniv.nl/interfac/cml/ssp/index.html
EEA (2004). EMEP/CORINAIR Emission Inventory Guidebook. European Environmental Agency.
Escobar, L. 2006. Indicadores sinteticos de calidad ambiental: un modelo general para grandes zonas urbanas.
Eure, XXXII, 96: 73-98.
FAO. 2008. FAOSTAT. In: http://www.fao.org
Huijbregts, M.A.J., Breedveld, L., Huppes, G., de Koning, A., van Oers, L., Suh, S. 2003. Normalization figures
for environmental life-cycle assessment The Netherlands (1997/1998), Western Europe (1995) and the
world (1990 and 1995). Journal of Cleaner Production, 11: 737-748.
IFA. 2008. International Fertilizer Industry Association. In: http://www.fertilizer.org
Jha, R., Murthy, K.V. 2003. An inverse global environmental Kuznets curve. Journal of Comparative
Economics, 31: 352-368.
Jollands, N., Lermit, J., Patterson, M. 2004. Aggregate eco-efficiency indices for New Zealand – a principal
component analysis. Journal of Environmental Management, 73: 293-305.
Judez, L. 1989. Tecnicas de analisis de datos multidimensionales: bases teoricas y aplicaciones en la agricultura.
MAPA, 301 pp.
Levitan, L. 2000. “How to” and “why”: assessing the enviro-social impacts of pesticides. Crop Protection, 19:
629-636.
MAPA. 2006. Hechos y cifras sobre la agricultura. Ministerio de Agricultura, Pesca y Alimentacion. Secretaria
General Tecnica, Madrid, 159 pp.
MAPA. 2008. Precios semanales en la cadena de comercializacion manzana. MAPA, 2 pp.
Marten, G.G. 1988. Productivity, stability, sustainability, equitability and autonomy as properties for
agroecosystems assessment. Agricultural Systems, 26: 291-316.
Neilsen, D., Neilsen, G.H. 2002. Efficient use of nitrogen and water in high-density apple orchards.
HortTechnology, 12: 19-25.
Rodriguez-Diaz, J.A., Camacho, E., Lopez-Luque, R., Perez-Urrestarazu, L. 2008. Benchmarking and
multivariate data analysis techniques for improving the efficiency of irrigation districts: An application in
Spain. Agricultural Systems, 96: 250-259.
Saaty T.L., 1990. How to make a decision: the analytic hierarchy process. European Journal of Operational
Research; 48: 9-26.
Sands, G.R., Podmore, T.H. 2000. A generalized environmental sustainability index for agricultural systems.
Agriculture, Ecosystems and Environment, 79: 29-41.
Sea Distances. 2008. Sea distances – Voyage calculator. In: http://e-ships.net
Soler-Rovira, J., Arroyo-Sanz, J.M. 2003. Spatial and temporal trends of environmental indicators relating to
nitrogen and phosphorous in Spain. In: Schnug, E. et al. (eds). Fertilizers in context with resource
management in agriculture. 14th International Symposium of Fertilizers. Hungary, pp: 321-329.
Soler-Rovira, J., Arroyo-Sanz, J.M. 2004. Organic nutrient management indicators at district level in Spain. In:
Hatch, D.J. et al. (eds). Controlling nitrogen flows and losses. Wageningen Academic Publishers, The
Netherlands, pp: 231-233.
Van den Berg, N.W., Dutilh, C.E., Huppes, H. 1995. Beginning LCA: A guide into environmental Life Cycle
Assessment, Center for Environmental Science (CML), Leiden.
ViaMichelin. 2008. ViaMichelin. In: http://www.viamichelin.es
Wackernagel, M., et al. 1999. National natural capital accounting with the ecological footprint concept.
Ecological Economics, 29: 375-390.

Proc. of the 6th Int. Conf. on LCA in the Agri-Food Sector, Zurich, November 12–14, 2008 page 142 of 414
Assessment of aggregated indicators of sustainability using PCA: the case of apple trade in Spain

Waterfootprint. 2008. Water footprint network. In: http://www.waterfootprint.org

Wurtenberger, L., Koellner, T, Binder, C. 2006. Virtual land use and agricultural trade: Estimating
environmental and socio-economic impacts. Ecological Economics, 57: 679-697.
Yu, C., et al. 1998. Effective dimensionality of environmental indicators: a principal component analysis with
bootstrap confidence intervals. Journal of Environmental Management, 53: 101-119.

Proc. of the 6th Int. Conf. on LCA in the Agri-Food Sector, Zurich, November 12–14, 2008 page 143 of 414